| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Polymer Sci. and Eng., Univ. of Massachusetts, Amherst, MA 01003 USA
b Chemistry Dep. and the Barnett Institute, Northeastern Univ., Boston, MA 02115 USA
srohr{at}iastate.edu
bx{at}pssi.umass.edu
srohr{at}iastate.edu
bx{at}pssi.umass.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMethods and materialsResults and discussionConclusionsREFERENCES |
|---|
Abbreviations: CP, cross-polarization • CPMAS, cross-polarization magic-angle spinning • CP/T1, cross-polarization spin-lattice relaxation time • CSA, chemical shift anisotropy • DPMAS, direct-polarization magic-angle spinning • HAs, humic acids • IHSS, International Humic Substances Society • NMR, nuclear magnetic resonance • PEMs, plant-extracted materials • TOSS, total sideband suppression • WHL water hyacinth leaf • WHR, water hyacinth root
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMethods and materialsResults and discussionConclusionsREFERENCES |
|---|
One of the most promising techniques for studying the chemical structure of humic substances is 13C NMR, (Preston, 1996). For soluble samples, such as humic or fulvic acids, solution NMR can be employed (Preston and Schnitzer, 1987; Thorn et al., 1989). However, the resonance lines are broad and low, necessitating much longer signal averaging times than are needed for solution NMR of small molecules with sharp lines; many hours to days are required if quantitative spectra are desired. In addition, solution NMR is not suitable for insoluble samples such as whole soil or humin. In this respect, solid-state NMR is more versatile. Its sensitivity is quite good due to the high sample concentration, and it is indeed the most widely used magnetic-resonance technique for soil analysis. The most popular 13C solid-state NMR technique used in studies of humic substances is CPMAS (Wilson, 1987). However, it is well known that this standard solid-state NMR technique has various problems with quantification (Fründ and Lüdemann, 1989; Kinchesh et al., 1995; Preston, 1996; Wershaw and Mikita, 1987; Wilson, 1987).
The most widely appreciated shortcoming of CPMAS NMR is the reduced cross-polarization (CP) efficiency for unprotonated carbons, mobile components, or regions with short TH1
(proton rotating-frame spin-lattice relaxation time). The shortening of TH1 by paramagnetic species such as Fe in HAs, to <5 ms in many cases (Pfeffer et al., 1984), causes protons to lose magnetization before complete transfer to unprotonated carbons is achieved. This problem is present at all magnetic-field strengths. The second well known factor that makes CPMAS spectra nonquantitative arises from magic-angle spinning sidebands, which reduce the intensity of the main line, the centerband. This problem occurs prominently when the rotation frequency is smaller than the chemical shift anisotropy (CSA), which is largest for sp2–hybridized carbons such as aromatic and carbonyl groups and increases proportional to the magnetic-field strength. While a high spinning rate can decrease the sidebands, it also interferes with CP (Axelson, 1985; Stejskal et al., 1977). The TOSS pulse sequence (Dixon, 1982; Dixon et al., 1982) reduces the sideband problem only in part, because the intensity suppressed in the sidebands reappears only partially in the centerband, and less so for the sp2–hybridized carbons with their large CSA (Axelson, 1985; Schmidt-Rohr and Spiess, 1994). The third problem in obtaining reliable spectra is the baseline distortion due to a dead-time at the start of detection. This is particularly serious for HAs due to their broad spectral lines and wide dispersion of chemical shifts.
Attempts have been made to improve the quantification by the use of ramped cross-polarization, which even under high-speed MAS establishes the Hartmann-Hahn matching condition necessary for CP (Metz et al., 1994, 1996; Peersen et al., 1993). An amplitude ramp on either of the radio frequency channels (1H or 13C) during the CP contact time improved the performance of CP experiments. This technique was recently applied in the study of humic substances (Cook et al., 1997; Cook and Langford, 1998), but the contact times needed according to studies of crystalline organic compounds (10 ms) are too long to avoid a significant distortion of peak intensities in HAs by TH1 relaxation (Metz et al., 1996).
An alternative that avoids most of the drawbacks of CPMAS is DPMAS at high rotation speeds. However, this technique, which has occasionally been employed in coal research (Maroto-Valer et al., 1996; Jurkiewicz and Maciel, 1995), has not found widespread use because for quantitative spectra, it requires the recycle delays between scans to be five times longer than the longest 13C spin-lattice relaxation time TC1. In noncrystalline solids, the longest TC1 is often of the order of 5 s to >200 s, so that the recycle delays of 5 TC1 are forbiddingly long; in addition, the determination of the long-time equilibrium magnetization necessary for measuring TC1 is extremely tedious. While paramagnetics generally shorten the relaxation times in humic materials, in some humic substances the longest TC1 is still of the order of tens of seconds (Kinchesh et al., 1995). In fact, crystalline components, which may have even longer relaxation times, have been found in humic substances (Hu et al., 2000). Consequently, for humic substances with potentially long TC1's it is impractical to use DPMAS alone to obtain quantitative 13C NMR spectra.
In this paper, we demonstrate that DPMAS combined with a TC1 correction obtained from CP/T1–TOSS spectra is a viable approach for quantitative NMR analysis of humic substances. We have previously used this approach to quantify polymer crystallinity (Hu and Schmidt-Rohr, 2000). By employing DPMAS, the CP problems are avoided. The sidebands can be reduced to an insignificant proportion by fast sample spinning: At 13 kHz spinning and a field strength of 7 Tesla, the largest sidebands, those of the aromatics, are suppressed to a total of <8% of the centerband (Herzfeld and Berger, 1980), and placed outside the region of the centerbands, so that they can be integrated easily. The baseline problem can be minimized by a Hahn spin echo (Hahn, 1950; Harris, 1983; Schmidt-Rohr & Spiess, 1994) before detection. The effects of incomplete relaxation are corrected by factors directly measured in CP/T1–TOSS experiments. The objectives of this study are (i) to assess the quantitative reliability of DPMAS corrected by CP/T1–TOSS for humic substances, PEMs, and even whole peat soil, (ii) to characterize humic substances of different origins and PEMs, and (iii) to compare the chemical composition information provided by NMR with structural models of HAs.
|
Methods and materials |
|---|
|
TOPABSTRACTINTRODUCTIONMethods and materialsResults and discussionConclusionsREFERENCES |
|---|
In our approach, recycle delays of 1.3 TC1 or even less can be used in DPMAS experiments, because we determine the signal fraction that has not relaxed within this time, using a CP/T1–TOSS experiment. This missing fraction is determined for each peak, and in the DPMAS spectrum the peaks are corrected for their missing fractions. With recycle delays of >1.3 TC1, the corrections are <25%.
Figure 1
shows the relatively simple NMR pulse sequences used in this approach. The DPMAS experiment, Fig. 1a, uses single-pulse excitation of 13C and a Hahn spin echo before detection to avoid baseline distortions due to dead-time problems. In the CP/T1–TOSS pulse sequence shown in Fig. 1b, after cross-polarization from 1H and before the TOSS 180°–pulse train, a + z/-z filter is applied so that the signal decays from full intensity towards zero as a result of TC1 relaxation and phase cycling (Torchia, 1978). Two CP/T1–TOSS spectra are run at two different TC1 filter times. The first has a TC1 filter time
, which is so short that virtually no TC1 relaxation occurs. The second has a filter time t±z equal to the recycle delay used in the DPMAS experiment. For example, the DPMAS spectrum of ARC HA was obtained with a recycle delay 5 s, and the second TC1 filter time of its CP/T1–TOSS spectrum was also set at
. The signal fraction remaining after the T1–filter is the same that will be missing in the DPMAS spectrum. This means that the smaller the remaining signal, the better, since the correction in the DPMAS spectrum will be smaller.
|
 | (1) |
) is the DPMAS peak area of the fully relaxed spectrum, and T1 is the 13C spin-lattice relaxation time. The relaxation function r(trecy) is exponential,
, for specific chemical sites. However, in HAs overlap of signals of similar groups in different environments can result in a non-exponential r(trecy). We therefore consider the most general case here.
|
 | (2) |
, and h(t±z) is the CP/T1–TOSS peak height after a filter time t±z. In our experiments, we chose
. Then,
 | (3) |
Substituting
for r(trecy) in Eq. [1], we obtain the corrected intensity:
 | (4) |
This procedure assumes that all components contributing to one peak have the same CP efficiency or the same TC1 relaxation time. This assumption is generally justified, since the CP efficiency for groups with identical chemical shifts (i.e., similar environment) is usually similar. Only for extended graphitic structures with few protons and therefore some large C–H distances and a wide range of T1 times would significant deviations be expected (Jurkiewicz and Maciel, 1995).
Experimental Procedure
Origin and Preparation of Humic Acids and Plant-Extracted Materials
Table 1
lists the origins and some analytical data of the samples studied. One peat soil (IHSS Pahokee peat), six peat HAs [German, Irish, Amherst, New Hampshire (NH), and New York (NY)], three commercial HAs (IHSS-LEON, ARC, and Aldrich), and three PEMs [from the brown alga Pilayella littoralis (L.) Kjellman (Ectocarpales), water hyacinth (Eichornia crassipes) leaf and root (WHL and WHR)] were investigated. The six Histosol HAs [German, Irish, Amherst, NH, and NY] were extracted from their corresponding surface-layer (0–20 cm) peats. The PEMs were obtained from plants following the HA extraction procedures (Ghabbour et al., 1994). The extraction and purification procedures were described in detail elsewhere (Ghabbour et al., 1994; Klute, 1986). Aldrich HA was re-extracted and purified in our laboratory before use.
|
For CP-TOSS, samples were packed in a 7-mm-diam. zirconia rotor with a Kel-F cap and run at 75 MHz on a Bruker MSL-300 spectrometer at a spinning speed of 4.5 kHz. The 1H 90° pulse length was 3.4 µs, the carbon 180° pulse was 6.4 µs. The contact time was 500 µs. The recycle delay was 1 s, and 4096 scans were recorded.
The CP/T1–TOSS was used to measure TC1 relaxation factor by modifying the CP–TOSS pulse sequence, see Fig. 1. After the contact time and before the TOSS 180°–pulse train, a +z/-z filter was applied so that the signal decays from full intensity to zero as a result of TC1 relaxation (Torchia, 1978). The number of scans was 2048 or 4096 with TC1 filters of 5 s or less; however, 1024 scans gave sufficient signal-to-noise ratio and were used with TC1 filters of 25 s.
Due to its good sensitivity, the CP/T1–TOSS experiment can also be run in the same 4-mm probe as the DPMAS, at a spinning speed of 4 to 6 kHz, which can make the experiments more straightforward for the spectroscopist. The feasibility of this approach was confirmed for the whole peat sample, which was only investigated in the 4-mm probe. Its CP–TOSS spectrum in Fig. 8 , obtained with 4096 scans, has a similar signal-to-noise ratio as the other spectra, showing that the smaller size of the 4-mm rotor does not reduce the sensitivity drastically.
|
Evaluation of Direct-Polarization Magic-Angle Spinning Spectra Corrected by Cross-Polarization Spin-Lattice Relaxation Time-Total Sideband Suppression
The following describes the steps used in the determination of the ideal fully relaxed 13C signal intensities.
Experimental Estimate of 13C T1
The CP/T1–TOSS experiment was used to estimate the 13C T1 relaxation times. The T1 filter time t±z was increased from a starting value of 2 s until the intensity of every peak in the second CP/T1–TOSS spectrum had decayed to less than half of its height in the reference spectrum (i.e., the spectrum obtained at
). This ensures that the correction factors in Eq. [4] are relatively small and thus minimizes the effects of noise and baseline distortion.
Acquisition of a Direct-Polarization Magic-Angle Spinning Spectrum
The spectra were obtained under the conditions described in the section Nuclear Magnetic Resonance Spectroscopy. The recycle delay time depended on the longest 13C T1 relaxation time in the given sample. If the longest 13C T1 of the sample as measured in the CP/T1–TOSS experiment was <2 s, as for samples with high ash content, the recycle delay was chosen as 5 TC1, which yields a completely relaxed DPMAS spectrum. Otherwise, the recycle delay was chosen equal to the T1 filter delay t±z in the corresponding CP/T1–TOSS experiment.
Deconvolution of Spectra
The Bruker Xedplot 2.2.0 deconvolution software was used to deconvolute a spectrum composed of overlapping peaks into individual bands (Pierce et al., 1990; Wershaw et al., 1996). The positions of the fitted bands were determined from the obvious peaks in the spectrum. Nevertheless, if one peak could not be fitted with a single band, another band was used to improve the fit. The bands were fitted with 100% Gaussian line shapes; Lorentzian line shapes have long tails and would lead to excessive overlap of neighboring bands. By measuring T2, we have found experimentally that the line-broadening is predominantly heterogeneous, which justifies Gaussian line-shapes.
Relaxation Correction Based on Two Cross-Polarization Spin Lattice Relaxation Time-Total Sideband Suppression Spectra
If the DPMAS spectrum was not the result of complete relaxation, two CP/T1–TOSS spectra were run. Then, Eq. [4] was used to correct the DPMAS spectrum.
Calculation of the Sideband Percentage of sp2 Carbons
Aromatic and carbonyl groups have the greatest CSA. At 13 kHz, the sidebands of a typical aromatic group are 7% of its centerband and 4% in the case of carbonyl groups. Thus, the sideband percentage of the two groups should be added to the centerband.
Ratio of sp2–C to sp3–C
The ratio of sp2–C to sp3–C was calculated from the corrected DPMAS spectrum area using Eq. [5]:
 | (5) |
Actually, there can be both sp2– and sp3–C in the 96 to 108 ppm region. However, for consistency and convenience in making comparisons, we include this range in the sp3–C.
The software of ACDs Spectrum Calculators was used to obtain chemical shifts for carbons in HA models. In the calculation of the ratio of sp2–C to sp3–C, the definition of Eq. [5] was used for the calculated chemical shifts as well.
Estimate of Elemental Composition from Nuclear Magnetic Resonance Data
In order to examine the reliability of the NMR technique, elemental compositions of %C, %H and %(O + N) were estimated from the NMR spectra obtained from DPMAS spectra corrected by CP/T1–TOSS, and then compared with the results from routine chemical analyses.
Calculation of %C, %H, and %(O + N)
From the band intensities at the various 13C chemical shifts (Breitmaier and Voelter, 1987; Cook et al., 1997; Malcolm, 1990; Stevenson, 1994), the elemental numbers of C, H, and O of different deconvoluted bands can be estimated. Table 2
lists the ranges used. As several different functional groups can contribute to a given range, the numbers of each element were generally obtained based on the average of the elemental numbers in those functional groups. For example, for the 145 to 162 ppm region, the main functional groups are aromatic C–O– and C–OH moieties. Thus, the C number is 1, the O number is 1, and the average of H is 0.5. Hence, the estimated average elemental composition of this chemical shift range is COH0.5. The functional groups in the 96 to 108 ppm range are assigned to anomeric carbons O–CH–O. The two oxygens are usually shared by other carbons, in C–O–CH–O–C linkages, and should therefore be only counted half, giving a elemental composition of CHO.
|
 | (6a) |
 | (6b) |
 | (6c) |
 | (6d) |
Estimate of Carbon Loss
Humic substances can contain significant amounts of inorganic and/or organic paramagnetic species. The dipolar couplings between the unpaired electrons and nearby nuclear spins can broaden the NMR signal of those nuclei beyond detectability. If these NMR-invisible species make up a significant fraction of the material and have a different composition than the sites more remote from the paramagnetics, the NMR analysis will be incorrect (Pfeffer et al., 1984, Preston et al., 1984; Preston and Newman, 1992; Skjemstad et al., 1994). We have determined the NMR carbon loss and assessed the spectral-intensity distortion by comparing the spectra of the same HA material before and after de-ashing.
|
Results and discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMethods and materialsResults and discussionConclusionsREFERENCES |
|---|
|
15 s). The recycle delays for the HAs with short relaxation times and for IHSS Florida peat were set to five times of their longest 13C TC1 in their DPMAS experiments. Thus, their fully relaxed spectra were obtained (Fig. 3)
and no CP/T1–TOSS corrections needed.
|
|
|
, respectively, which gives slightly improved agreement with chemical analysis.
|
Comparison of Elemental Compositions Estimated by Nuclear Magnetic Resonance and from Chemical Analysis
The %C, %H, and %(N + O) from NMR calculation and chemical analysis are compared in Fig. 6
. Under ideal conditions, if the two methods are in complete agreement, all the dots in Fig. 6 should be on the straight lines y = x. Actually, the dots in Fig. 6 are close to or on these lines, showing that NMR is consistent with elemental analysis. Only the NMR H content shows some scatter and seems to be systematically somewhat too high. This indicates a slight overestimate of the H content of some of the chemical groups in Table 2. The good agreement of the NMR and elemental analysis is not trivial. For instance, significantly inconsistent results were obtained in cases, for example, for Pilayella PEM and Irish HA, where the recycle delays were not set long enough; within a time equal to this recycle delay, the intensity in the CP/T1–TOSS filter experiments relaxed to no less than half, which resulted in large correction errors. These unreliable data were excluded from the further analysis. In summary, DPMAS with CP/T1–TOSS correction is a reliable technique if the second filter time of CP/T1–TOSS is long enough to make the intensity relax to less than half of the initial value.
|
|
Comparison of Humic Acids of Different Origins, of Plant-Extracted Materials, and of Peat
The CP–TOSS spectra indicate qualitatively differences in composition between the various HAs, PEMs, and IHSS Florida peat (Fig. 8). All the soil HAs are similar, except Florida HA and, to some extent, NY HA. The soil HAs exhibit distinct peaks at 31 ppm, 55 ppm, 75 ppm, 130 ppm, and 175 ppm. NY HA is very similar to Florida HA in the sp2–carbon region, both having two broad peaks without a peak around 150 ppm (Fig. 3, Fig. 8). The peat spectrum is quite similar to NY HA. Compared with soil HAs, Pilayella and WHL PEMs have a much higher percentage of sp3 carbons. WHR PEM has only three major peaks, indicating a relatively simple composition of this PEM. The IHSS-LEON, ARC, and Aldrich HA spectra are also relatively simple, with a sharp peak in the aliphatic region and a broad 100 to 180 ppm region, characteristic of coal HAs.
Figures 9 and 10 display the quantitative results obtained from the DPMAS spectra corrected by CP/T1–TOSS. Each spectrum has been partitioned into nine regions corresponding to relatively distinct peaks in the spectra of Fig. 3 and Fig. 8. They represent ketone–quinone–aldehyde, carboxyl–ester–quinone, phenolic, aromatic, complex aromatic–anomeric, carbohydrate–ether, methoxy–methyne, complex aliphatic, and simple aliphatic carbon concentrations. The anomeric peak, while not always clearly resolved, has been confirmed in Amherst HA based on its 1H chemical shift (J. Mao et al., unpublished data, 2000).
|
|
Comparison with Previous Solid-State Nuclear Magnetic Resonance of Humic Acids
Solid-state NMR has been used extensively in attempts to quantify the composition of HAs and other humic substances (Preston, 1996; Mahieu et al., 1999). Our results show that the CP spectra in the literature were clearly nonquantitative. The most striking and unequivocal difference is in the CO signals, which were underestimated by a factor of 
2 in the CP spectra. This is shown by comparison with CPMAS results for peat and Leonardite HAs in the literature (Ayuso et al., 1997), see Table 3. This significant difference, which agrees with findings by Golchin et al. (1997a)(1997 b) who compared CP with DPMAS for a few samples, also reduced the sp2/sp3 ratios estimated from CPMAS spectra in the literature. In addition, there are other deviations of the literature CP data from our DP results. The aliphatic-carbon content obtained from the CPMAS spectra of peat HA according to Ayuso et al. (1997) was unusually high (29.1%), and their Leonardite HA CP spectra had a carbohydrate signal much higher than in any of the three old humic acids characterized here.
The systematic underestimation of the CO carbons in CP spectra is due to their long distance from the nearest protons, which makes them cross-polarize poorly. Ramped CP does not alleviate this problem significantly, because the varying spin-lock field strength means that unprotonated sites are Hartmann-Hahn matched only for a small fraction of the total spin lock time. There is insufficient time for complete transfer on the time scale during which decay due to TH1 is negligible. For a typical humic-acid TH of 4 ms, the signals decays by 22% within 1 ms of spin lock, which is barely long enough to cross polarize the unprotonated carbon sites fully. In addition, the sidebands of the sp2–carbons, which detract from the centerband intensity, have also been consistently underestimated.
The erroneously low CO content and reduced sp2– to sp3–carbon ratio have led to incorrect conclusions, such as statements that humic acids are predominantly aliphatic. Our results show the necessity of a significant revision of data and models that used previous CPMAS NMR results quantitatively. Only comparisons between samples, that is, relative changes of peak intensities, are still valid (Wilson, 1987). Even then, the sensitivity of the CP condition to small changes in field strengths makes sp2–carbon peak intensities unreliable.
Comparison with Humic Acid Model Structures
Over the years, various models for HAs have been proposed. Based on our NMR quantification of HAs, we can now test these models. The right hand side of Fig. 9 displays the compositions of several model structures according to the nine spectral ranges, obtained based on chemical-shift calculations using the ACDs Spectrum Calculators software. The resulting spectra are arranged in an order that matches that of the experimental spectra to some extent.
Steelink (1985) proposed a tetramer HA model containing aromatic rings, phenols, and quinones linked by aliphatic units with many OH groups. The COOH groups in this model are linked exclusively to aliphatic groups. The composition in terms of the various sp2–C matches that of the older soil HAs (NY and Florida), but the structure is lacking OCH3 or CH groups in the 50 to 60 ppm range, as well as anomeric carbons. These should be added to produce an acceptable model of HA structure.
Modifying Steelink's model (1985), Jansen et al. (1996) proposed a building block of HAs which has seven chiral centers and thus 128 stereoisomers. Instead of quinones, it exhibits ketones or aldehydes. Again, the sp2–C composition matches that of NY and Florida HAs well, but simple aliphatics (signals below 35 ppm) as well as anomeric carbons are missing. Nevertheless, among the models tested here, those of Steelink and Jansen et al. represent two of the closest approximations to the composition of the soil HAs, in particular in the sp2–C region, and may be a good starting point for model refinement.
Based on the idea that the structure of HAs was generated by a combination of four predominantly aromatic building blocks, namely a dimer formed by the coupling of two lignin-derived oxidation products, a phenol–amino acid complex, a hydroxyquinone, and a C6–C3 structural unit of lignin, Stevenson (1994) developed a model which he considered to "contain many requirements for a `typical' soil humic acid," by adding units such as a condensed aromatic ring to the four building blocks. In Stevenson's model, the COOH groups are mostly attached to aromatic rings, which dominate the structure. Amino acid sidegroups, which are only vaguely defined in the model, were not included in the analysis here. The concentration of aliphatic components in the core structure of Stevenson's model is clearly lower than in the experimental spectra of soil HAs. Most of the 96 to 108 ppm signal in this model is due to complex aromatic structures and therefore shown in black like the other aromatics.
Schulten and Schnitzer's complex model (1993) reflects the results from CPMAS 13C-NMR (which we have shown to be unreliable), analytical pyrolysis, and oxidative degradation data. The structures are aromatic rings linked by long-chain alkyl structures and have many COOH and OH groups on both the aromatic rings and aliphatic side chains. In spite of the model's complexity, the composition does not match the soil HAs particularly well; however, it has some resemblance to the coal-extracted (IHSS-LEON, ARC, and Aldrich) HAs as shown in Fig. 10.
Dragunov's model (Kononova, 1966) suggested that the main structure of soil HA at least partially consists of aromatic rings of the di- or trihydroxyphenol-type bridged by –O–, –(CH2)n–, –NH–, –N– and contains COOH, OH and quinone-type linkages. This main structure is linked with proteinaceous and carbohydrate residues through covalent bonds. The resulting composition has similarities with the experimental results for the extract from WH root, but does not match the soil HAs.
Leenheer and coworkers (Averett et al., 1989) proposed three basically similar models of fulvic acids. These models all contain two aromatic rings and one tetrahydrofuran ring as main blocks, and some methyl-terminated sidegroups. Ketones but no quinones are included. As shown, the structure has a resemblance to the WH leaf extract. It is interesting to note that of all the models, this one exhibits the sp3–C structure which resembles that of the soil HAs most closely. After reduction of the number of COO groups and addition of phenolic groups as well as OCH3 and amino acid groups, the match to the soil HAs would be quite acceptable.
Figure 10 includes an old model by Fuchs (Stevenson, 1994), which was derived from results for coal HAs. It consists of a condensed ring system to which COOH and OH groups are attached. In fact, it reproduces several of the features of the three coal-extracted HAs. The main discrepancy is the absence of simple aliphatics from the model. Here, the 96 to 108 ppm signal arises from aromatic structures and is therefore shown in black. Also shown in Fig. 10 is Flaig's model (Stevenson, 1994), which consists of many linked aromatic, phenolic, or quinonic rings but contains only few COOH and aliphatic groups. Except for having too many phenolic and too few COO groups, it resembles the ARC coal-extracted HA very closely.
The sp2/sp3 carbon ratios of the different HA models in the literature were calculated according to Eq. [5] and compared with the experimental data, where it can be obtained nearly assumption-free by integrating two quite distinct regions of the NMR spectra. The sp2/sp3 carbon ratios of all the model, except Flaig's, range from 1.19 to 3.56 (Fig. 11) . The ratio of sp2 to sp3 of soil HAs range from 1.11 to 2.83 while those of coal HAs varies from 1.58 to 3.44. The wide range sp2/sp3 ratios of HAs used in this study cannot be represented by a single model in the literature. Flaig's, Fuchs's, Stevenson's, and Steelink's models contain predominantly aromatic carbons, which makes them similar to the commercial, coal-derived HAs in our study. There are more aliphatics in Jansen et al.'s, Dragunov's, Leenheer's, and Schulten and Schnitzer's model.
|
|
Conclusions |
|---|
|
TOPABSTRACTINTRODUCTIONMethods and materialsResults and discussionConclusionsREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
Received for publication June 22, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMethods and materialsResults and discussionConclusionsREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. Farenhorst Importance of Soil Organic Matter Fractions in Soil-Landscape and Regional Assessments of Pesticide Sorption and Leaching in Soil Soil Sci. Soc. Am. J., April 19, 2006; 70(3): 1005 - 1012. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. J. Dougherty, R. J. Smernik, and D. J. Chittleborough Application of Spin Counting to the Solid-State 31P NMR Analysis of Pasture Soils with Varying Phosphorus Content Soil Sci. Soc. Am. J., October 27, 2005; 69(6): 2058 - 2070. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Chilom, S. D. Kohl, and J. A. Rice The Influence of Lipids on the Energetics of Uptake of Polycyclic Aromatic Hydrocarbons by Natural Organic Matter J. Environ. Qual., May 11, 2005; 34(3): 1055 - 1062. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
